ML Machine Learning Basics Part 8

Описание: 21. Sample & Population

Population = the whole group. Sample = a small part of it.
Example: All students in a school = population. Students in one class = sample.
ML often uses samples because it’s difficult to collect data for the entire population.

22. Train/Test Split

We divide data into training and testing parts.
Training data: teaches the model.
Testing data: checks if the model learned correctly.
Example: Practice questions = training. Final exam = testing.

23. Cross Validation

Instead of one test, we split data into many small parts and test multiple times.
Example: A student gives 5 small practice tests instead of one big exam, to check real understanding.
ML uses cross validation for fair evaluation.

24. Normal Distribution

It’s a bell-shaped curve most data follows.
Example: In a class, most students score around average, few get very high or very low marks.
ML and statistics use this curve often to understand data.

25. Cost Function (Loss Function)

A cost function tells how wrong the model’s predictions are.
Example: If a student guessed answers, the teacher calculates how many marks he lost.
ML models try to reduce the cost function — meaning, they try to make fewer mistakes.